Piano-Playing Robotic Arm

摘要

This project explores the intersection of robotics and technology with music. The study looks specifically at the expressive aspects of a human performer and how to translate and represent that in a robotic system. By analyzing live performances of multiple performers and talking with professional performers, the team built an understanding of human gesture in performance. This was demonstrated by the creation of a robotic piano-playing system, including an industrial arm and custom-built hand. To fully understand the interpretation and performance of music, the team implemented a set of machine learning techniques, which included training a recurrent neural network (RNN) to analyze audio signals and reproduce musical input.